QF1-002 - Comparison of three automated quantitative myocardial perfusion analyses versus invasive fractional flow reserve in swine

Thursday, February 13, 2020

2:48 PM – 2:51 PM

Location: Sabal/Sago (Meeting Room in Exhibit Hall)

Background: In the absence of microvascular disease, invasive fractional flow reserve (FFR) should correspond to the myocardial blood flow (MBF) ratio (MBFratio) between an ischemic area and a normal area. Automated quantitative perfusion analysis by three methods was compared with invasive FFR in swine.

Methods: Ten swine underwent one or more computer-controlled partial or total occlusions of the left anterior descending artery (LAD) yielding a total of 25 experiments. FFR was recorded simultaneously with a dual bolus injection protocol. The dual echo, dual sequence perfusion acquisition sequence was used. Three software packages estimated pixel-wise MBF (ml/min/g). Method 1 (previously validated) and Method 2 (prototype commercial software based on Method 1) both used the dual bolus data for the arterial input function (AIF). Method 3 used the dual echo, dual sequence data for the AIF. All three methods used the same high contrast myocardial perfusion images. Each experiment yielded an ischemic and normal MBFratio.

Results: Ischemic zone FFR values averaged 0.74 +/- 0.13. CMR MBFratio correlated well with FFR for Methods 1-3 (y=0.950x-0.044, r=0.87; y=0.988x-0.088, r=0.83; y=1.248-0.314, r=0.82 respectively where y=CMR MBFratio and x=FFR). Using a threshold of 0.8 for FFR and MBFratio, the three methods had good diagnostic accuracy (Figure 1). AUC was not significantly different in pairwise comparisons. Absolute MBF correlated well (r=0.83 for 2 vs 1; r=0.86 for 3 vs 1; r=0.88 for 3 vs 2).

Conclusion: In carefully controlled, simultaneous experiments, all three pixel-wise perfusion mapping methods reliably distinguished hemodynamically significant coronary stenoses and predicted invasive FFR values.

Presenting Author(s)

AB

Andrew J. Bradley, MD

Clinical Fellow
National Heart, Lung, and Blood Institute, National Institutes of Health

Disclosure: Disclosure information not submitted.

Primary Author(s)

AB

Andrew J. Bradley, MD

Clinical Fellow
National Heart, Lung, and Blood Institute, National Institutes of Health

Disclosure: Disclosure information not submitted.

Co-Author(s)

DG

Daniel W. Groves, MD

University of Colorado School of Medicine

Disclosure: Disclosure information not submitted.
MB

Mitchel Benovoy, PhD

Director
Circle Cardiovascular Imaging Inc.

Disclosure: Disclosure information not submitted.
SY

Sarah Yang, BA

Nih

Disclosure: Disclosure information not submitted.
SK

Shawn Kozlov, BSc

Virginia Maryland College of Veterinary Medicine

Disclosure: Disclosure information not submitted.
JT

Joni Taylor, BSc

Surgery Manager
National Heart, Lung, and Blood Institute, National Institutes of Health

Disclosure: Disclosure information not submitted.
AS

Arlene Sirajuddin, MD

Radiologist
National Heart, Lung, and Blood Institute, National Institutes of Health

Disclosure: I do not have any relevant financial / non-financial relationships with any proprietary interests.
LH

Li-Yueh Hsu

Staff Scientist
National Heart, Lung, and Blood Institute, National Institutes of Health

Disclosure: Disclosure information not submitted.
Andrew E. Arai, MD, FSCMR

Chief
National Institutes of Health

Disclosure: Circle CVI (Individual(s) Involved: Self, Products/Services: Cardiac MRI software): Received grants or funds in support of research of related company products.; Corstem (Individual(s) Involved: Self, Products/Services: Cardiac MRI software): Patents and Invention Reports